Centre for Law, Economics and Society
CLES
Faculty of Laws, UCL
Director: Dr Ioannis Lianos
CLES Working Paper Series
4/2012
Market Power in Online Search and SocialNetworking: a Matter of Two-Sided Markets
Florence Thépot
September 2012
1
Market Power in Online Search and SocialNetworking: a Matter of Two-Sided Markets.
Florence Thépot1
The industries of online search and social networking are characterised by high
market shares held by a very limited number of actors. Google holds 85% of the
market for Internet search engines in terms of traffic, and has been maintaining its
market share at this level since 2008. Eight years after its creation, the socialnetworking website Facebook has 800 million active users in 2012, accounting for
about 65% of the market share of social-networking websites, in terms of registered
users. As advertised-based media, search engines and social-networking websites
have the characteristics of a special type of market known as two-sided markets or
platforms. Two-sided markets are platforms which have two distinct user groups
providing each other with network benefits. The platforms enable the user groups to
minimise the transaction costs they would have otherwise incurred, of searching for
each other and interacting. Economics of two-sided markets cannot be ignored by
competition authorities in assessing market power in the search and social
networking industries. The paper provides a framework for defining the relevant
market and for assessing market power in the industries of online search and socialnetworking websites, focusing on the current leaders Google and Facebook. It is
argued that online-search and social-networking websites may exert competitive
constraints on each other, as both operate in the relevant market for ‘monetization of
user’s information by online advertising’.
1. Introduction
The fast-evolving markets for online search and social networking are characterised
by high market shares held by a very limited number of competitors. Eight years after
its creation, the social-networking website Facebook has 800 million active users in
2012, accounting for about 65% of the market share of social-networking websites in
the US.2 Google holds around 88% of the global market for Internet searches, and
has been maintaining its market share at this level since 2008.3 On-going EU and US
investigations show that dominance in these markets may raise competition
concerns: for instance, Google has recently been investigated by the European
Commission for allegation of abuse of its dominant position relating to alleged
1
2
3
Florence Thépot is a PhD candidate at University College London, and a fellow of the Centre for
Law, Economics and Society. A draft of this paper was presented at the 19th CLaSF workshop on
th
‘Competition Law, Regulation and the Media’, City Law School, Thursday 26 January 2012. I am
grateful to the organizers and the participants for their valuable comments.
Kallas, ‘Top 10 Social Networking Sites and Forums 2008-2012.’ Available at:
http://www.dreamgrow.com/top-10-social-networking-sites-by-market-share-of-visits-august-2012/
(retrieved 09/10/2012).
KarmaSnack,
‘Search
Engine
Market
Share.’
Available
at:
http://www.karmasnack.com/about/search-engine-market-share/ (retrieved 09/10/2012). For the
evolution of search engine market share since 2008, see Andromida, ‘Search Engine Market Share
Statistics.’ Available at: http://andromida.hubpages.com/hub/search-engine-market-share (retrieved
11/10/2011).
2
unfavourable treatment of search services companies.4 The Federal Trade
Commission (FTC) opened a formal investigation against Google concerning similar
allegations.5
While opening the probe against Google, the European Commission acknowledged
the difficulty of establishing dominance in the Internet search market due to the
particular nature of Google’s business.6 Indeed, search engines and socialnetworking websites have the characteristics of a special type of market known as
two-sided markets or platforms. Two-sided markets are platforms which have two
distinct user groups providing each other with network benefits. The platforms
provide a meeting place to two groups or parties, and enable them to minimise the
transaction costs they would have otherwise incurred, of searching for each other
and interacting.7 To do so, such platforms may engage in matchmaking or building
audiences.8
Google, as other search engine portals, acts as an intermediary between advertisers
and users. Google provides a service that attracts users who form an audience that
in turn attracts the advertisers. As with other advertising-supported media, Google
earns all of its revenues from advertisers.9 Social-networking websites, such as
Facebook, engage in building an audience and are mostly advertising-based, while
their core function for the users is also to be an exchange platform.
Common features of two-sided markets are that each group on one side of the
platform tends to realize more value when there are more users on the other side.
For example, the interest of using social-networking websites for users increases the
more groups of people use it too as it saves the cost of interacting with these people
on bilateral basis.10 Two-sided markets are characterised by indirect network
externalities that the platform helps internalise. Prices and profits are linked on the
two sides and each side of the platform exerts some constraint on the other.11 This
has practical implications in terms of market definition and assessment of market
power: for example, the market has to be defined in relation to the other side of the
market; the feedback effects between the two markets need to be accounted for in
4
European Commission, ‘Joaquín Almunia Vice President of the European Commission responsible
for Competition Policy Statement of VP Almunia on the Google antitrust investigation.
SPEECH/12/372, 21/12/2012.
5
Sara Forden, ‘FTC Said Poised to Finish Google Antitrust Probe in Weeks’, Bloomsberg, 30/08/2012.
6
See also Almunia, Competition in Digital Media and the Internet. Speech delivered at UCL, 7 July
2010. For a video recording of the speech, see: http://www.ucl.ac.uk/news/newsarticles/1007/10070802 (retrieved 10/10/2011).
7
D. S. Evans and R. Schmalensee, 'Markets with Two-Sided Platforms' (2008) 1 Issues in
Competition Law and Policy 667.
8
D. S. Evans and R. Schmalensee, 'Industrial Organization of Markets with Two-Sided Platforms’
(2007) 3 Competition Policy International 152.
9
D. S. Evans and R. Schmalensee (n 7) 671.
10
In the case of advertising-based media, the advertisers value the platform the more users there are,
but the users may not value the platform if there are more advertisers. The platform internalises the
negative externality suffered by the users by 'paying them' to use the service, ie. by offering free
content. Most search and social-networking web portals are indeed free of use. See Evans and
Schmalensee (n 8) 155-156.
11
D. Evans, 'The Antitrust Economics of Two-Sided Markets' (2003) AEI-Brookings Joint Center For
Regulatory Studies 49.
3
terms of prices and profits, and the barriers to entry have to be assessed differently.12
Ignoring the specificities of such markets may yield erroneous conclusions in
competition analysis.13
In addition to being two-sided platforms, social-networking and search websites have
common and special features such as being advertised and running as web-based
media. Moreover, both industries seem to be currently led by ‘winner-takes-all’
companies, respectively Facebook and Google. This makes it relevant to analyse
both industries alongside each other, with their special features demanding extra
care in the assessment of market power.
In this paper, it is argued that the economics of two-sided markets cannot be ignored
by competition authorities when dealing with search and social networking industries.
After evaluating the economic foundations that are useful in this case, the different
characteristics of the search and social-networking websites as platforms for twosided markets will be established, focusing on the current market leaders Google and
Facebook. The paper will provide a framework for assessing market power in the
markets for search and social-networking websites, encompassing an assessment of
relevant Commission decisions.14 The issue of search bias, predominant in the
current investigation on Google’s alleged abuse of dominance, as well as the issue of
overall consumer welfare are outside the scope of this paper.
2. Search and social-networking websites: example of two-sided platforms.
2.1. Two-sided markets: some economics
The economics of two-sided markets appeared with the works of among others,
Caillaud and Jullien,15 Rochet and Tirole,16 Evans and Schmalensee,17 and
Armstrong.18 Two-sided markets comprise an intermediary whose role is to act as a
platform between two distinct markets. The two different markets produce indirect
network effects on each other, and each side is linked with a feedback effect. A
common example of two-sided markets would be a heterosexual dating club. Men
value the club only if women come to the club, and vice versa. The platform, here the
dating club, enables both group of customers, men and women, to interact and
search for each other. The platform helps economize the searching cost incurred
12
13
14
15
16
17
18
Ibid.
For an account of the fallacies that arise from using the one-sided market logic in assessing twosided markets, see: J. Wright, One-Sided Logic in Two-Sided Markets (2003) AEI-Brookings Joint
Center For Regulatory Studies.
Merger cases: COMP/M.4731 Google/DoubleClick, 11 March 2008 and COMP/M.5727 –
Microsoft/Yahoo! Search business, 18 February 2010.
B. Caillaud and B. Jullien, 'Competing cybermediaries' (2001) 45 European Economic Review 797;
'Chicken & egg: competition among intermediation service providers' (2003) 34 RAND Journal of
Economics 309.
J.C. Rochet and J. Tirole, ‘Platform Competition in Two-sided Markets’ (2003) 1 Journal of the
European Economic Association 990
D.S. Evans & R. Schmalensee (n 7 and 8).
M. Armstrong 'Competition in Two-Sided Markets' (2006) 37 RAND Journal of Economics 668.
4
compared with if the men and women were to operate on a bilateral basis. The
platform also helps get the balance right between the two groups of users via its
pricing strategy. Charging the women a lower entry fee than the men is an example
of price structure commonly adopted in dating clubs.
A two-sided market exists if two groups of agents interact via an intermediary, and if
the decisions of each group impact the outcomes on the two sides, typically through
an externality, such as a network effect.19 In economic terms, there is a two-sided
market if two groups of consumers are linked by an externality and if they cannot
internalise it on a bilateral basis due to high transaction costs.20 In two-sided markets,
the price-cost mark-up does not just depend on the elasticity of demand and marginal
cost, but also on the elasticity of demand on the other side, as well as on the pricecost mark-up charged.21
These conditions make two-sided markets different from other markets even if in
most markets, parties interact via an intermediary too. In these markets, such as the
wheat market, the price structure does not matter in terms of optimal outcome. A tax
levied on a buyer will have the same effect as a tax levied on a seller. In contrast, the
price structure in a dating club matters: a lower price for women than for men brings
a higher outcome, which equally matters as the price level. Similarly, there might not
be such interdependence between the two sides of the intermediary in a one-sided
market. The profit of a farmer selling once wheat to an intermediary - e.g. a grocery may not be linked to the profit of the grocery selling the good. In that case, the
conditions for a two-sided market are not fulfilled in spite of the existence of an
intermediary.22
The intermediary or platform plays an essential role in two-sided markets: that of
internalising the externalities, and diminishing transaction costs between the two
groups of users. To do so, platforms may engage in matchmaking users, such as in
exchange platforms. Another type of two-sided market is advertising-based media.
Such media acts as a platform between users and advertisers. They engage in
building an audience to make available to a pool of advertisers. Web portals,
including online search and social networking websites are example of advertisedbased media. As operating with two markets, the platform must follow a different
profit-maximising approach than a traditional business. The platform has to account
for the demand of each side of the platform, and the effect the demand of one side
has on the other side. Similarly, the platform must take into account the costs
attributed to each side, as well as the costs of running the platform.23
Taking into account two-sided market characteristics is of fundamental importance
when undertaking a competitive assessment. Not doing so, a competitive
19
20
21
22
23
M. Rysman, 'The Economics of Two-Sided Markets' (2009) 23 Journal of Economic Perspectives
125.
D. S. Evans and R. Schmalensee (n 8) 154.
M. Rysman (n 19) 129.
Ibid, 126.
D. S. Evans, 'Two-Sided Market Definition', in Market Definition in Antitrust: Theory and Case
Studies (chapter XII, ABA Section of Antitrust Law, 2009)7.
5
assessment might fall into various fallacies, identified by Wright.24 A first erroneous
conclusion induced by using a one-sided approach to a two-sided market is that a
price structure must reflect relative costs, in the logic that users must pay to cover the
costs. However, in the example of the night club, the men do not pay in relation to the
cost of using the club, but in relation to the surplus they derive from an additional
woman on the other side of the platform. The price does not necessarily reflect the
cost of use, but it can still be a competitive market. Related to this issue, a high pricecost margin does not necessarily indicate the existence of market power. Similarly, a
below-cost pricing does not necessarily stem from a predatory pricing strategy.
Another one-sided fallacy is to think that more competition can impact on the price
structure. It is likely that a monopolist night club will price the same way as many
clubs competing fiercely with each other: higher price for men than for women. More
competition will certainly drive the price level down, but is unlikely to bring a more
efficient price structure. Also, it would be incorrect to consider the price structure as a
cross-subsidy between the two sides of the market. This would ignore the indirect
network effect at stake: in the night club example, the participation of men depends
on the participation of women. As a result, the revenue to the club associated with an
additional woman generates the revenue associated with an additional man, which
amounts to the total revenue gained by the club. Each customer therefore triggers
revenue that covers more than the incremental cost. In contrast, a cross-subsidy
implies that one side produces a gain below the incremental cost of its participation,
and that the whole platform would be better off without their participation, in terms of
revenue, which is not the case in terms of two-sided markets.25
2.2. Google and Facebook, platforms of two-sided markets.
2.2.1. Online search and its current market leader, Google.
Over the last 15 years, the business of online search has grown very rapidly in line
with the general development of the Internet, if not faster. Finding a piece of
information among the immense quantity of data available on the Internet has
become tremendously important if not the core tool of Internet users. At the end of
the 1990s, Google emerged in the online search landscape, with a reputation for
producing more relevant results than the existing search engines, its algorithm being
based both on text matching and on ‘reputation’, a proxy for the quality of the web
page.26 Just two years after its launch, Google accounted for 7 million searches per
day. Today its revenue amounts to $12.21 billion for the quarter ending 30 June
2012,27 and it holds around 85% of the market for online search. The use of the
search engine is free to users. Online search companies fund their system with
advertising revenues. They sell companies advertising space alongside search
results, tailored to the keywords of the search being conducted. On Google, two
24
25
26
27
J. Wright (n 13).
Ibid 2-9.
S. Brin and L. Page, 'The Anatomy of a Large-Scale Hypertextual Search Engine' (1998) Computer
Science Dep’t, Stanford Univ. (1998), http://infolab.stanford.edu/pub/papers/ google.pdf
Google,
Google
Announces
Second
Quarter
2012
Financial
Results.
Available
at:http://investor.google.com/earnings/2012/Q2_google_earnings.html (retrieved 09/10/2012).
6
types of result appear as the outcome of a search query: the list of websites as
‘organic’ results, and the corresponding list of ads that have been paid for by
advertisers. In other words, Google attracts traffic with its search functionality and
sells the traffic’s ‘attention’ to advertisers as a way to generate its revenues.
2.2.2. Social networking and its current market leader, Facebook.
Social networking websites provide an array of means for users to interact and
socialize with one another: the facilities offered include traditional internet
communication such as messaging and email, as well as the possibility to share
pictures, videos and files. In addition, social networking websites are used for
blogging, discussion groups, among other interaction means.28 Facebook, created in
2004, is a market leader in this industry, with a market share of 65% of social
networking users in 2011. Other examples include websites such as Twitter (microblogging), LinkedIn (professional network), or MySpace (music). Facebook derives its
revenues from advertising. Unlike other advertising-based media, users themselves
create the content that attracts traffic, which in turns attracts the advertisers.
Facebook’s platform also supports other web-firms via the various applications it
hosts.29
2.2.3. Why are Google and Facebook two-sided platforms?
Google and Facebook are both advertising-based media. As with many web portals,
advertising is used to fund the system offered to users for free. Both operate as the
intermediary between two groups of agents: in one side, the web surfers that make
search queries or use the social networking content. On the other side, they operate
with a wide range of advertisers.
In economics terms, Google and Facebook intermediate the transactions between
the users and advertisers helping internalise the network externalities. They enable
exchanges to be realised, which may not otherwise occur.30 To some extent they
provide the ‘liquidity’ of the market by increasing the volume of buyers and sellers
that can achieve mutually profitable transactions. Absent a certain amount of liquidity,
the market may be too narrow and unsustainable.31
Google: Search-based advertising
28
29
30
31
D R Gnyawali, W Fan, J Penner, 'Competitive Actions and Dynamics in the Digital Age: An Empirical
Investigation of Social Net' (2010) 21 Information Systems Research 595-596 in S. Weber Waller,
Antitrust and Social Networking (2012) North Carolina Law Review 6.
D. S. Evans, 'Antitrust Issues Raised by the Emerging Global Internet Economy' (2008) 102
Northwestern University Law Review 291.
D. S. Evans & R. Schmalensee (n 8) 152.
D. S. Evans, 'The Economics of the Online Advertising Industry' (2008) 7 Review of Network
Economics 372.
7
Google attracts users with its search tool, as well as with other functionalities such as
email, maps, videos, and as the owner of YouTube. We focus here on the search
functionality offered by Google. In turn, Google's users, attracted by the content,
attract advertisers to whom Google sells advertising space. More advertisers on one
side and more users on the other side may increase the likelihood of beneficial
matches between the users and the advertisers, which is the case for exchange
platforms.32 If verified, this illustrates the indirect network effect between the two
sides of the platform. If this is the case, the nature of this indirect network effect
needs more attention.
The advertisers value the user side of the platform as users constitute an audience
pool for the advertisements. The larger the platform is, the more interesting it gets for
the advertisers as there are higher chances that purchases will be undertaken by the
users. It is not as clear that users value the other side of the platform as much as the
advertisers do. Users may value the online search more if it provides advertisements
that are relevant to the query, providing that they wish to purchase something.33 In
that situation the question is whether more advertisers on one side bring more
relevant advertisements to the users. This seems to be true, especially for narrowly
advertised markets, for which a larger advertiser base makes a difference to the
users’ value. If there is no purchase intention, the users may stop using the website if
there are too many advertisements. However, the platform may ‘internalise’ the cost
of viewing ads by offering the service for free. In other words, advertisements
disturbance might be seen as the non-price fee borne by the users.34 Moreover,
increased advertising enables Google to fund the refinement and development of the
organic search functionality, and helps Google provide more free functionalities on
top of the search engine (email, video and picture storage and so on).35
Google also enables people to look for web-based businesses which appear in the
list of organic search results. These businesses, which include other publishers, do
not pay for their ads to be displayed and benefit from the platform. Other web
publishers also subscribe to Google's advertising services. They allow Google for
some space to insert its ads, for which the publishers receive a part of Google's
revenue.36 As such Google interacts with different actors that are themselves twosided platforms: one example being social-networking websites.
Facebook: non-search advertising
Facebook attracts users with its social-networking system. It does not create content,
but provides the framework for the creation of content by the users themselves
(although the applications and other functionalities also constitute content). The
content created by the users constitutes a pool of data that is used to attract the
advertisers. The advertisers can use Facebook for their advertisements in different
32
33
34
35
36
Ibid.
D. S. Evans (n 31) 293.
D. S. Evans & R. Schmalensee (n 7) 670-671.
K. Laudadio Devine, 'Preserving Competition in Multi-Sided Innovative Markets: How Do you Solve a
Problem Like Google?' (2008) 10 North Carolina Journal of Law & Techology 83.
D. S. Evans (n 31) 296.
8
ways: by creating a page for the brand advertised, by purchasing advertising space
which will be displayed on Facebook's website and by buying 'sponsored stories' as
well as embedding Facebook functionalities on the business' own website. When
purchasing advertising space, the advertisers can select the audience to target,
according to data available on age, location and so on. Facebook's functionality
enables advertisers and potential buyers to increase the chance of matching their
interest mutually and profitably.
Facebook also offers to display advertisements by the way of 'sponsored stories'.
Sponsored stories are conveyed by users which interact on Facebook with the
business's page or application: the story comprises the action of the user in relation
to the business such as 'Martin is having coffee at Starbucks', as well as the picture
brand and link to the business page. The picture constitutes the advertisement for
which the business pays for. The advertisement is then conveyed by an individual
user to all of his contacts.37 From the advertisers' perspective this is a very powerful
way of leveraging the networks of its customers/subscribers. Sponsored stories also
entail recommendations by users that convey stories. The question of benefits to the
user is similar to search-based advertising. The larger the pool of advertisers and
users, the greater is the chance of a successful transaction. The users might be
disturbed by the display of advertisements, in spite of their being relevant to their
profile. In the case of the sponsored stories, the user's experience of the
advertisements might be quite different. In that case, the advertisement becomes
user content, sent by the user to their contacts. Users might value sponsored stories
the same way they value non-sponsored stories published by their contacts. How
they value this relates to the concept of network effects characterising all networks:
each user values the system the more users there are, and the more they produce
content. While the network effect can be limited by capacity constraints, or
congestion, such limits do not seem to affect Facebook, which is still seeing its
number of registered users grow. The settings of the system allow each user to
select the users they wish to interact with, and the amount of information they want to
see from each user, which helps reduce the potential cost to users of a too large and
intense network.
Both platforms face a ‘chicken and egg’ issue in operating with two groups of
consumers.38 Advertisers have no incentive to buy advertising space if the users are
not yet on board on the other side. The other side is the source of revenue attached
to the transaction. Similarly, an absence of advertisers will decrease the value of the
platform to the users, if this implies that the users have to pay for use. This is the
characteristic of indirect network effects between markets whose participation affects
each other’s participation. More participants on one side induce more participants to
join the other side of the platform. This positive feedback effect may well affect the
market structure and the type of competition which then prevails. Large platforms
usually see this pattern, as they are deemed to bring higher benefits to advertisers
and users. The market for online advertising is populated by numerous platforms, but
characterised by a few very large market leaders in their categories (online search,
37
38
Facebook, Sponsored Stories for Marketplace. Available at:
https://www.facebook.com/ads/stories/SponsoredStoriesGuide_Oct2011.pdf (retrieved 13/01/2012).
B. Caillaud et B. Jullien (n 15).
9
social networking). Each category of online-based advertising is characterised by the
domination of one platform, which can vary across countries.39 For example, Google
is the leader in western countries, while Baidu holds most of the Chinese online
search market.40
In summary, Google and Facebook are example of platforms of two-sided markets,
characterised by indirect network effects which they help internalise, so as to bring
value to both advertisers and users. The market structure in which both market
leaders operate can be related to the existence of two-sided markets. In the case of
Facebook, this is combined with the usual direct network effect that features each
social-networking. In order to assess market power, the subtleties linked to these
characteristics must be accounted for in order to avoid erroneous conclusions.
3. A framework for assessing market power.
We now turn to the actual analysis that must be realised by competition authorities in
order to assess market power in such markets. The analysis encompasses several
steps: the first stage is the market definition that is used to calculate market shares in
the relevant market. Then the market share must be put in relation with the potential
competitive constraints that determine market power, including the contestability of
the market, the existence of barriers to entry and exit for new competitors, as well as
the possibility of consumers to switch to another service. All these steps will be
analysed in the light of two-sided markets economics. We will first examine market
definition issues (3.1), before turning to competitive constraints (3.2) in the market for
social-networking and online search.
3.1. Market definition
Competition authorities start their enquiry with market definition in the context of
abuse of dominance41 and merger control procedures42. A first aim is to delineate the
scope of the enquiry, in order to limit it to a manageable set of relevant products and
companies. The relevant market includes all the products with which the product at
stake may compete, with which there is a sufficient degree of substitutability.
Substitution, which can come from the demand as well as from the supply side of the
product, is a key concept in this respect. If from the consumer's perspective a
competitor’s good is a very good substitute for the product at stake, the demand
elasticity is likely to be high, and therefore the company will not manage to raise the
price sustainably, as the consumer has the possibility to switch to another product.
This rationale underpins the test used by most competition authorities to define the
relevant market, namely the Small but Significant and Non-transitory Increase in
39
40
41
42
D. S. Evans (n 31) 298.
Simon Montlake, ‘China's Qihoo 360 Takes Aim At Baidu's Search 'Monopoly'’, Forbes, 31/08/2012
Article 102, Consolidated version of the Treaty on the Functioning of the European Union (TFEU),
[2010] OJ C 83/01.
Council Regulation (EC) No 139/2004 of 20 January 2004 on the control of concentrations between
undertakings (the EC Merger Regulation) [2004]OJ L 24.
10
Price or SSNIP test. In this test, we look at the effect of a price increase of a 5 to
10% on the profit of the industry. If this price increase triggers a loss due to a lower
demand, it means that the product has available substitutes that need to be included
in the relevant market. If the price increase induces increased revenue, there is a
lack of available substitute elsewhere, which shows that this is a relevant market that
is ‘worth monopolizing’ by a hypothetical monopolist.43 The SSNIP test is used both
in abuse of dominance and merger control cases.44 The larger the relevant market,
the less likely dominance will be found in that market.
The particularities of two-sided markets have several practical impacts with respect to
market definition. First, there is more than one market to examine, as there are two
sides to the platform. Secondly, the tools used in traditional markets must be used
with care, so as to account for the feedback effect between the two sides. The
elasticities of demand of the two sides are intertwined, therefore a hypothetical
increase in price on one side cannot be analysed in isolation to its effect on the other
side. We will first analyse the definition issues in the two sides taken separately
before turning to the tools used to define the relevant market of the platform as a
whole.
3.1.1. Market definition on one side: the functionality offered to users
(online search, social-networking)
We will consider the potential definition issues of the user side of the platform, first of
the online search websites, and then of the social-networking websites. We will
illustrate by focusing on the market leaders, Google and Facebook.
3.1.1.1.
Online search.
From the user side perspective, Google provides a functionality for online searches,
together with offering various functionalities, such as map services, email and
messenger tools, as well as video service since its acquisition of YouTube. In this
paper we focus our analysis on the search functionality offered by Google. A search
engine is a “computer program that searches databases and Internet sites for the
documents containing keywords specified by a user.” 45 In its decision
Microsoft/Yahoo! Search, the European Commission describes it as a
[A] tool designed to search for information on the Internet. It consists of a
search box in which queries can be typed. The search results of a given query
are then usually presented in a ranked list of results. The information
searched for may consist of text (including news), maps, images, videos or
other types of content.46
43
44
45
46
S. Bishop and M. Walker, Economics of E.C. Competition Law: Concepts, Application and
Measurement (Sweet & Maxwell, 2009).
In one-sided logic a number of caveats of the SSNIP test are identified: the cellophane fallacy,
teethless fallacy etc, which do not touch upon the themes of this paper.
Business dictionary online, retrieved 13/01/2012
DG COMP, M.5727, para. 30.
11
From the user viewpoint, the 'search engine' offered by Google has as substitute all
similar functionalities available to users on the web offered by other providers, such
as Bing, Yahoo, Baidu, Ask and so on. Those general search engines must be
distinguished from vertical search engines which focus
on specific segments of online content such as for example legal, medical, or
travel search engines. Contrary to general Internet search engines, which
index large portions of the Internet through a web crawler, vertical search
engines typically use a focused crawler that indexes only web pages that are
relevant to a pre-defined topic or set of topics.47
The Commission also held that general search engines must be distinguished from
search tools integrated into websites, which enable users to search for content within
a website.
The search engine industry is highly concentrated. Looking at market shares, based
on web traffic (views of a website), Google clearly leads its competitors. 48
Interestingly, Facebook appears in some market share rankings for online search,
with 1.43% of market share for the US market.49 Similar to many websites, Facebook
integrates a search tool for users to find contacts or content. In addition, there is now
a directory for contacts, places and businesses that is publicly available. It can be
seen as a way for Facebook 'to challenge third-party search engines as the preferred
way to find information about businesses. This is because advertisers are willing to
pay higher prices to appear when users are making decisions about which local
business or restaurant to visit.'50 For now it seems that it would not be appropriate to
place Facebook's search tool in the same market as general search engines, but the
dynamics of such industry might well change this in the future.
Another aspect is whether there is a substitute available to users in the 'brick and
mortar' market. This expression refers to the 'offline' version of a service offered on
the Internet. Online search is very specific and inherent to the development of the
Internet and to the information that is available. Search engine can be seen as
providing a one-stop shop to services that may also be available offline, such as
search engine rather than library for information research, searching for flights online
rather than going to a travel agency and so on. This is of course not specific to online
search, as the whole economy is increasingly populated by online counterparts of
businesses. Therefore, from the user perspective there is no substitute to engine
search in the 'brick and mortar' market.
Even if from the user's viewpoint, the 'product' seems to be all general search
engines, from an antitrust point of view, it is incomplete to consider Google as
47
48
49
50
Ibid, para. 31.
Karmasnack (n 3).
Ibid.
Constine, To Complement Its Search Engine, Facebook Launches People, Pages, and Places
Directories. Inside Facebook, available at: http://www.insidefacebook.com/2011/09/28/facebooklaunches-people-pages-and-places-directories-to-challenge-search-engines/ (retrieved 13/01/2012).
12
exclusively operating in the relevant market for online search. It ignores the other
side of the platform, and the interrelations between both. In its decision, the
Commission described the online search product, but correctly considered the
relevant market to be in the field of advertising.51
3.1.1.2.
Social-networking.
Social networking web portals, previously defined as providers of an array of means
for users to interact and socialize with one another - messaging, email, pictures,
videos, file sharing, blogging, discussion groups - can encompass many different
type of web portals. Under this definition of social networking, Facebook has a
leading position in the category, as it offers all the services described. Other websites
offer a similar range of functionalities such as Google+, LinkedIn or Twitter. These
websites are focused on the identity of people, and the interaction among them.
Other social-networking websites may be more centred on specific interests, such as
music or news, but there are no borders that can be drawn between different
categories as most social-networking websites also enable people to connect via
their specific interest.
As for substitutability, we leave aside for now the fact that the use of one website is
not exclusive from the use of another website. We theoretically ask the question of
choices available to consumers, should the access to one of the website be
constrained. This might include the introduction of an access fee, or stricter
registration requirement. What would happen if Facebook were to restrain access to
a category of people, as was seen when it originally only provided access to the
students of some American colleges? This is more than theoretical and does not
really fit the SSNIP test, as not relating to a small increase in price, but this does give
an idea of the concept of substitution we are using here. Some websites can fall
under the definition of social-networking websites: Groupon, offering discounts,
Amazon, the retail website, as well as YouTube. However, they appear to be only
partial substitutes to social networking such as Facebook, as they only provide one
part of its functionality. This view was adopted in the US case LiveUniverse v.
Myspace,52 in which the District Court distinguished dating website from general
social-networking websites. The Court concluded that MySpace operated in the
relevant market for Internet-based social networking website. If their definition of
social-networking seems appropriate, we don't subscribe to how the market is
defined as not accounting for the two-sides of the market.
51
52
COMP, M.5727. We will examine below the details of the market definition.
LiveUniverse v. Myspace CV 06-6994 AHM, 2007 WL 6865852 (C.D. Cal. June 5, 2007). 'Internet
connectivity services are not reasonable substitutes, because their primary function is simply to give
users the ability to access the Internet. As to online dating sites, although they do have similar
“organic, interactive qualities” to social networking websites, their dominant function and purpose is
to enable users to meet potential dates. Online dating sites are not reasonable substitutes for social
networking websites, because the latter websites have significantly more functions and appeal than
do online dating sites. For example, social networking websites are used to get in touch with old
friends and to keep current friends informed about what’s new and exciting. Although social
networking websites may also be used for dating, if MySpace suddenly were to shut down, its
members would not fill the social void by turning to online dating sites. Instead, they would likely set
up profiles on a different social networking website.'
13
Social-networking websites such as Facebook do not have counterparts in the 'offline'
life. Many 'brick and mortar' businesses provide places where people gather and
exchange according to their identity and interest. Those places may operate as a
platform between different groups of users, and help internalise indirect network
effect, as do social networking websites. However, social networking websites
provide a very distinct experience to users, specific to their online nature. The means
of communications are numerous and online-related. On top of this, Facebook
provides various types of communication means that can apply to different segments:
professional networking, hobbies, specific interest, charity management etc. In other
words, Facebook has a potential to be used in relation with every dimension of
peoples’ lives, which can make it a large one-stop platform for users. Such equivalent
cannot be found in the 'brick and mortar' market.
3.1.1.3.
On the other side: advertisers
Both Facebook and Google earn their revenues from the other side of the platform
they operate, namely advertising. Online advertising-based companies have already
been the focus of competition decisions, in which market definition in the field was
approached. A first set of issues in this respect relates to whether online and offline
advertising, such as newspaper advertising, belong to the same relevant market. In
the context of Google and search-based advertising, the question is also whether
non-sponsored advertising can compete with sponsored ads.
Online and offline advertising pursue similar objectives, namely, to 'inform, persuade,
remind or motivate consumers by delivering information, rhetoric, and/or imagery to
those consumers.'53 Every means of reaching the targeted audience has its own
characteristics and offers different functionalities to both advertisers and users. The
question is thus from an advertiser point of view whether the offline environment can
be a substitute to the online? To what extent do both sides exert competitive
constraints on each other? Would it be sufficient to group them in the same relevant
market?
We will first consider the potential constraint exerted by online advertising on offline
advertising. Before the rise of the Internet, online advertising had very little potential
to constrain the more traditional and long-established means of advertising, such as
radio, TV newspapers or ‘snail mail’. The audience reachable online was too narrow,
thus the advertisers had very little incentive to spend budgets towards online
advertising. Nowadays, the Internet's audience is global, offering a great potential for
advertisers. Internet advertising represented 10.6% of all advertising expenditures in
the US, and was expected to reach 13.6% by 2012.54 A shift in expenditure from
newspapers to online advertising was expected to occur from 2007 to 2012,
53
54
J.D Ratliff and D. L. Rubinfeld, 'Online Advertising: Defining Relevant Markets' (2010) 6 Journal of
Competition Law and Economics 12.
Norweb, ‘Online Ad Spending Slows but Grabs Market Share.’ Available at:
http://www.norwebpromotions.com/online-ad-spending-slows-but-grabs-market-share/commentpage-1/#comment-1278 (retrieved 14/01/2012).
14
accompanying a decrease in the readership of those newspapers.55 These figures
show that a shift from offline to online stems from a shift of the audience from offline
to online media. This does not indicate on the substitutability between online and
offline advertising, in the event an offline media raises its advertising prices. The
constraint on price has been recognised in a US court decision, in which the alleged
newspaper advertising market was deemed to be too narrowly defined because it did
not include non-print media advertising.56
As for the constraint exerted by the online market on the offline market, it seems that
both will continue to compete. The decrease in newspaper readership is only one
aspect of offline advertising as it includes also TV and radio; this part of offline
advertising will evolve as being no longer a competitive constraint to online
advertising. Some empirical studies approached the difficult question of
substitutability. In the study 'Advertising Bans and the Substitutability of Online and
Offline Advertising’, the authors analysed whether Internet advertising is more
effective in places that prohibit advertisements for alcohol displayed on billboards.57
By doing this, they are able to compare the effectiveness of online ads when there is
no offline competition, with its effectiveness when offline competition is possible. The
results show that offline advertising constrains the effectiveness of online advertising.
The authors acknowledge that their study does not comprise any pricing dimension,
but this still provides a robust result in terms of substitutability.
Another other study from the same authors, entitled 'Search Engine Advertising:
Substitution when Pricing Ads to Context' focused on search-based advertising,
examining how offline advertising impacts the price of search-based advertising. In
some US states, lawyers cannot contact by post a client victim of an injury before a
certain period after the accident. During this time, competition between online and
offline legal advertising can be deemed to be absent. The authors compared the
prices for online advertising associated with the keyword “brain injury attorney Baton
Rouge” with the price of an advertising related to the search “divorce attorney Baton
Rouge”. In the states with solicitation restrictions, the model shows that personal
injury keywords were relatively more expensive to the price of other legal keywords,
in comparison with the price gap between the same keywords in states without
regulation prohibiting such contact. These results may not be fully consistent to
define the relevant market for antitrust purposes, as the SSNIP test analyses the
response to a small price increase, not to an access ban to the market. Therefore
this study highlights the potential constraint effect of offline advertising on online
advertising, as the price of online advertising is higher in the absence of offline
advertising.58
55
J. D. Ratliff and D. L. Rubinfeld (n 53) 13.
Berlyn Inc. v. Gazette Newspapers, 73 Fed. Appx. 576, 582-83 (4th Cir. 2003).
57
A. Goldfarb and C. Tucker, ‘Advertising bans and the substitutability of online and offline advertising.’
(2011) 48 Journal of Marketing Research 207-227.
58
A. Goldfarb and C. Tucker, 'Substitution between offline and online advertising markets' (2011) 7
Journal of Law and Economics 37.
56
15
The European Commission considered whether offline and online advertising could
be in the same relevant market in various decisions.59 In all its decisions, the
Commission found a separate market for the online advertising. In Telia/Telenor, the
parties submitted that from the demand-side perspective, companies did not
exclusively advertise online, and that the advertising activity also encompassed the
offline channels. The Commission rejected this statement, stating that this was
related to the advertising strategy pursued by companies, to advertise via various
channels. With Google/DoubleClick, the Commission distinguished the offline from
online advertising markets because of the different characteristics of advertising
through these channels: differences in potential scope of targeted audience and in
the assessment of advertisements’ effectiveness; different pricing mechanism as in
online advertising prices connect the cost and the reach of the advertisement, which
is not the case in offline mechanism. In establishing a distinction between online and
offline advertising, the Commission failed to account for the potential competitive
constraints and related substitution between offline and online advertising. However it
is unlikely that the authorities will reverse their findings as such approach has been
adopted consistently over past decisions.
We now consider another market definition issue, within the market of onlineadvertising: whether non-search advertising may constrain search-advertising.
Search advertisements appear aside the organic search results, and are associated
with keywords that the advertiser has selected. In contrast, non-search
advertisements appear on a website without being related to a specific keyword.
From the advertisers’ viewpoint, search-advertising may seem to produce a more
targeted advertising, and increase the relevance of advertising. Search engines can
save keywords searched by a same user in the recent past. For instance Google now
uses interest-based advertising, aiming to show ads relevant to the websites visited
previously.60 This enable to gather information about the needs of the user, based on
the keyword search and on previous websites visited.
Non-search advertising may gather a great amount of information on viewers, which
allow to achieve a targeting of equivalent efficiency. A website, on which non-search
advertisements are displayed, attracts a segmented type of viewers with a specific
interest, which is likely to correspond to a particular gender, age category.
Geographic information can be inferred from the viewer’s IP address. 61 Websites
which require users to register can collect even more detailed valuable information.
Facebook is obviously one of those websites which gather a large amount of
information, potentially on every dimension of a viewer's identity, activities and
interests. From an advertiser's perspective, search-advertisements and non-search
advertisements can reach a very targeted audience. They probably target the
audience differently as a search engine has highly relevant information on the
59
60
61
Case IV/JV.1 – Telia/Telenor/Schibstedt, 27 May 1998, Case IV/M.1439 – Telia/Telenor, 3 October
1999 and Case IV/M.0048 – Vodafone/Vivendi/Canal Plus, 20 July 2000; COMP/M.4731
Google/DoubleClick, 11 March 2008 and COMP/M.5727 – Microsoft/Yahoo! Search business, 18
February 2010.
http://www.google.com/ads/preferences/html/about.html retrieved 15/01/2012
J. D. Ratliff and D. L. Rubinfeld (n 53) 10.
16
searcher’s current needs, and a social-networking website provides detailed
information about the user's identity and interest.62
With Google/DoubleClick and Microsoft/Yahoo! Search business the Commission
approached
such
issues
without
reaching
definite
conclusions.
In
Google/DoubleClick, the market investigation showed that the different forms of
online advertisements can constitute substitutes for the advertiser, as differences of
technicalities and aims (brand awareness can be created by both types) between
both tend to diminish.63 The Commission also considered the supply-side
substitutability to conclude that both channels of advertising were not likely to be part
of the same relevant market. A publisher cannot replace space sold for non-search
by space sold for search-ads, and vice versa. A publisher can integrate a search tool
on its website and share profits with the search engine provider, but this does not
fulfil the conditions for a supply-side substitution. Moreover,
[W]hen a publisher decides to allocate a given space on a web page to a nonsearch (e.g.display) ad, this would not be substitutable with a "search
generated" advertising space, since the latter only appears on the page
generated by the search query entered by the user.64
In the corresponding decision in the US, the Federal Trade Commission (FTC) stated
that search-advertising and non-search advertising could not belong to the same
relevant market:
[T]he evidence in this case shows that the advertising space sold by search
engines is not a substitute for space sold directly or indirectly by publishers or
vice versa. Or, to put it in terms of merger analysis, the evidence shows that
the sale of search advertising does not operate as a significant constraint on
the prices or quality of other online advertising sold directly or indirectly by
publishers or vice versa.65
If the demand-side substitution is verified and the supply-side is not, still we can find
a significant competitive constraint between the prices of search and non-search
advertisements. The differences of technicalities originating from the differences of
type of supply do not seem to be a crucial determinant of a relevant market as long
as advertisers are ready to use them interchangeably in case of a small price
increase. The current evolution seems to confirm this tendency.66 Defining the search
and non-search advertising as constituting one market has a significant implication:
Google and Facebook potentially constrain each other, on the advertising side of the
platform they operate. Market power of each of them would then drastically shrink
under such definition.
62
63
64
65
66
Ibid 18.
COMP/M.4731 Google/DoubleClick, 11 March 2008, para 52.
Ibid para. 54.
Federal Trade Commission, Statement of Federal Trade
Google/DoubleClick, (FTC File No. 071-0170) 3 and 7
http://www.ftc.gov/os/caselist/0710170/071220statement.pdf.
D Ratliff and D. L. Rubinfeld (n 53) 18.
17
Commission
(December
Concerning
20, 2007)
3.2. Relevant market of the platform as a whole
Now that we examined market definition issues and their implications on each side of
the platforms, we will turn to the question of the relevant market in which Google and
Facebook, as platforms for two-sided markets operate.
The discussions developed above were necessary but insufficient in order to
conclude as to the relevant market. Adopting a one-sided logic for a market definition
in the case of two-sided markets is erroneous as it ignores that the platform operates
with two-markets, whose prices and profits are linked. Economic tools usually used in
competition enquiry are designed for the definition of single-sided markets, which
may trigger an incorrect market definition in the case of two-sided markets. Some
literature specifically analyses the caveat of the SSNIP test.67 An application of the
SSNIP test on the advertising-side of the platform ran by a search-engine will
account for competing forces exerted by other publishers on the advertising space it
sells. This relates to the discussion above. An increase in the price of the
advertisements has a demand and a revenue effect: the increase diverts some
advertisers from the search website to another type of publisher, while producing a
gain in revenue from a higher price. The loss in advertising demand incurred has an
impact on the user side. Users may want to switch to another search engine, which
offers more advertising supply. In turn, the loss of users further decreases the
demand of advertisers on the other side, as their advertisements now reach a
narrower audience. Therefore, a price increase realised in the SSNIP test can be
deemed profitable on the advertising side, but non-profitable if accounting for the two
sides. Suggestions as how to adjust the test, at least in theory, are available in the
economic literature. Noel and Evans provide a two-sided extension of the SSNIP test
that involves starting with the platform as product and adding its closest substitute.68
The price considered would be the weighted average of the price charged in two
sides. At each step, prices across sides and across platforms would need to be
optimised. However this can be quite complicated in practice, especially if more than
two sides are accounted for.69 In our situation, relying on the concept of price raises
difficulties as one side uses the platform for free. We could easily imagine non-pricing
elements that would divert the demand to another platform, but this might prevent the
definition to rely on economic and numerical tools.
The critical loss analysis is a method used to apply the SSNIP test in a simpler way,
depending on the data available.
It compares “Critical Loss” (CL) --the percentage loss in quantity of a
hypothetical monopolist's products that would be exactly enough to make an
X percent price increase in the price of all of its products unprofitable--to
“Actual Loss” (AL) --the predicted percentage loss in quantity that the
67
68
69
M. D. Noel and D. S. Evans, 'Defining Markets That Involve Multi-Sided Platform Businesses: An
Empirical Framework With an Application to Google’s Purchase of DoubleClick' (2007) AEIBrookings Joint Center For Regulatory Studies, Working Paper; L. Filistrucchi, 'A SSNIP Test for
Two-sided Markets: The Case of Media' (2008) Net Institute, Working paper.
Ibid.
M. D. Noel and D. S. Evans, ‘The analysis of mergers that involve multisided platform businesses’
(2008) 4 Journal of Competition Law & Economics 669-78.
18
monopolist would suffer if it did increase prices on all of its products by X
percent. A relevant market is found when Actual Loss equals Critical Loss for
a hypothetical monopolist of the given set of products in the proposed
antitrust market. If Actual Loss exceeds Critical Loss, the relevant market is
expanded to include more substitutes. Otherwise, it is contracted.70
Noel and Evans conducted an empirical study so as to assess the potential bias in
market definition using this tool with a one-sided rather than a two-sided logic in the
context of the Google/DoubleClick merger decision in the US. They conclude that
one-sided logic can lead to material errors in competition assessment. They also
advise practitioners to avoid formalistic and systematic market definition in two-sided
markets.71 In the case of the markets under consideration, the absence of a pricing
element on one side of the market imposes to move away from too formalistic
assessment.
Once the pitfalls of market definition are accounted for, we will turn to suggesting
possible market definition. The relevant approach might be to identify the group of
customers served by the platform, its likely rivals and then identify the various
businesses that serve these customers. The notion of substitution must be part of the
analysis, even though it does not rely on formalistic method underpinning the SSNIP
test.72
Both types of platforms interact with multiple businesses, but we accounted for the
sides that are common to both: the user side and the advertising side of the platform.
For each side, we assessed the competitive constraints and specific definition issues.
We turn to examine the different possible way of defining the relevant markets.
As developed previously, it would be incorrect to define the platforms as the
functionality they offer, namely online search and social-networking. Similarly, the
advertising side of the platform cannot constitute a relevant market in itself as
ignoring the user side of the platforms. In case of a merger control case, however,
the market definition may confine to the advertising side if it is where the merging
companies overlap, as it is the case in Google/DoubleClick.73
Another way to look at market definition would be to combine both sides: Google
would operate in the market for online-advertised based search engine, while
Facebook would be active in the market for online-advertised based social
networking website. However we have seen that the advertising side of both
Facebook and Google are likely substitutes. Therefore it would seem difficult to apply
a clear-cut distinction on competitive forces within online-advertising.
Finally, some interesting approach to market definition stems from how those
platforms monetize their operations as it gives information on ‘who the customers
70
71
72
73
Ibid 668.
M.D. Noel and D. S. Evans (n 69) 40.
Filistrucchi considers conditions under which the 1-sided logic of the SSNIP test can be applied to 2sided markets (n 67).
COMP/M.4731 Google/DoubleClick, 11 March 2008.
19
are, whether there is competition, and whether the absence or potential absence of
competition is a result of business acumen or anticompetitive conduct.’74 Both
Facebook and Google monetize their operations in the form of advertising revenue.
In the functionality we are focusing on, much of the operation consists in gathering
and pooling information on users, relevant to the other side of the platform and to the
platform as a whole.75 Therefore, it would be useful to consider the relevant market to
be in the area of 'monetization of users’ information to advertisers'.76 Competition
authorities may seem unlikely define the relevant market as broadly as suggested
here. However, this definition can help the authorities move apart from technology
and technicalities considerations used in market definition, so as to focus on
competition constraints, accounting for both sides of the platforms.
3.3. Contestability of the markets
In this section, let us assume that the relevant market is defined so as Facebook and
Google have lead market shares in their respective markets or industries. This is
likely to be the case in antitrust decisions for the time being. In that scenario antitrust
concerns need to consider the contestability of the market in order to conclude
whether they actually have market power. The assessment of market shares in
isolation of barriers to entry; consumers lock-in etc. can yield erroneous conclusions.
Market share calculation is only one of the elements used to assess market power.
3.3.1. Barriers to entry
A first element to assess is the barriers to entry to potential competitors. A market in
which high profits are made has a potential to attract new businesses who want to
get their share of the high profits. The barriers to entry in the search advertising
industry seem high, not just because of the cost of developing the algorithm and
underlying technology, but because of the cost of getting users on the platform. The
indirect network effects are strong and therefore as each side of the market grows it
gets more difficult for new entrants to compete. New entrants do not benefit from the
same scope of indirect effects as the incumbents.77 An entrant might be able to
design a better algorithm and compete with Google, however Google’s current lead
position has been maintained over the last years, and dominance is not likely to be
questioned in the on-going investigation.
The technology and resources to enter into the social-networking industry seem
widely available. However the barriers to entry are high due to network effects of two
types: on the user side, Facebook benefit from a strong direct network effect, as
74
75
76
77
‘Intel, Apple, Google, Microsoft, and Facebook: Observations on Antitrust and the High-Tech Sector’
ederal Trade Commission before the ABA Antitrust
Section , November 2010.
‘The true product Facebook brings to the “market” is not its technology, but the social information
about, and access to, its vast user base.’ in C. Butts, ‘The Microsoft Case 10 Years Later: Antitrust
and New Leading "New Economy" Firms’ (2010) Northwestern Journal of Technology and
Intellectual Property 290.
S.Weber Waller (n 28) 13.
K. Laudadio Devine (n 35) 88-89; N. Zingales, presentation on Market Definition and Search Bias in
th
Online Search and Advertising, 19 Competition Law Scholars Forum, January 2012.
20
more users trigger more users to join the platform. Facebook also enjoys indirect
effect characterising two-sided markets as more users on one side induce more
advertisers to join, with a positive feedback effect. However the social-networking
industry is moving fast. Until a few years ago MySpace was the leader of the
industry, also benefiting from network effects. We may see the emergence of a new
social-networking company that would manage to find a powerful concept or
technology to revert the tendency of Facebook’s increasing position. One of the
characteristics of the industries based on high innovation cost is that companies
compete for the market rather than in the market. Therefore it is useful to assess
dominance across time rather than at a precise point in time.
Search engines, (eg. Google) vertically integrate with the acquisition of content
provider websites (eg. YouTube) that compete with other content providers such as
social networks. Social-networking websites with high market shares, such as
Facebook may hold a significant competitive advantage with having highly detailed
and valuable information on users. Therefore, in the market for online advertising, if
Google and Facebook are able to compete with each other, this would significantly
reduce their market power on the online advertising market
3.3.2. Consumer lock-in.
Users on one side and advertisers on the other side do not use one but several
websites, which is called multi-homing.78 Users may use different social-networks or
search engines and are likely not to constrain themselves to using one. Many
Facebook users also frequent other social-networking websites, as they do not
necessarily focus on the same segment. The use of several search engines may be
less frequent amongst users.
On the advertisers side, companies are even more likely to use several and
differentiated platforms to advertise. This may have positive effects for competitors
with smaller market share that may benefit from the multi-homing strategies of
advertisers.
Another specificity of these industries is the non-durability of choices. Unlike the
choice of a computer or a mobile phone, a decision to visit a search engine or to
register to a social-networking need not be durable as it is free to users. As a result
‘lock-in is much less likely, particularly if multi-homing is possible.’79 While
‘competition is one click away’, as Google says,80 it seems that people remain loyal
to the search engine they use. This is a matter of users’ preference or ‘laziness’
which does not relate to an effective consumer lock-in. As for Facebook, the closure
of one’s account seems simply done, but it will be only deactivated to allow users to
re-join at any time. Permanently deleting one’s account entails a more difficult
procedure. The deletion is active only after two weeks of deactivation, during which
78
79
80
J.C. Rochet and J. Tirole (n 16).
R. Schmalensee, ‘Jeffrey Rohlfs’ 1974 Model of Facebook: An Introduction’ (2011) MIT Sloan
School Working Paper 4893-11, 10.
http://www.google.com/publicpolicy/issues/competition.html, retrieved 19/01/2012.
21
the user must not log in again, nor interact with its account on third-party websites.81
This can be perceived as a switching cost to users as the procedure is not as simple
as just 'clicking away'.
Advertisers are also 'one click away' from other publishers. However, the network
and positive feedback effects reduce the incentives to quit a very large platform, as
they represent potential higher revenues than smaller platforms. An advertiser who
appears in the top of the organic search result may have further reduced incentives
to stop paying for advertising space if there is a doubt on the reliability of the
algorithm. Google adjusts the algorithm to counter 'search engine optimization' firms
that seek to optimize the rank of apparition of companies in the organic result by
'tweaking keywords and web design.'82 The search bias is a core question in the
current investigation, however there is no evidence that Google manipulates organic
search results at the moment.
4. Conclusions.
In online-search and social-networking platforms, the assessment of market power
must account for specificities of two-sided markets, as well as to issues specific to
each side of the platform. We concluded that for the user side of the platforms,
Facebook and Google do not have ‘offline’ equivalents. On the advertiser side, the
offline market has a potential to constrain the online market but this approach does
not seem to be adopted by competition authorities. Within the online-advertising, the
increasing substitutability between the non-search ads and search ads implies that
Facebook and Google could theoretically belong to the same market. This would
considerably reduce their market power and reduce antitrust concerns. It fits the idea
that the relevant market may be constituted by all platforms attracting advertisedbased revenues by monetising relevant information they hold on users. This
approach shows how focusing on competitive constraints and substitution can help
abstract from technology considerations for market definition purposes. Market power
depends also on the existence of barriers to entry that must be put in perspective
with the importance of network effects characterising those industries.
81
82
http://www.wikihow.com/Permanently-Delete-a-Facebook-Account retrieved 19/01/2012
J. Battelle, The Search: How Google and its Rivals Rewrote the Rules of Business and Transformed
our Culture (Nicholas Brearley Publishing, 2005) in K. Laudadio Devine (n 35) 85.
22